Why these puzzles feel hard, and what to do about it
You are not bad at this. You are running out of room.
There is a specific feeling that comes with a hard matrix puzzle: you find a rule, start checking it against the grid, and by the third cell you have lost the rule. You go back, find it again, and lose your place instead. After a few rounds of this the conclusion arrives that you are simply not good at these.
That conclusion is almost always wrong. What you are running into is working memory, and it is a hard limit that has very little to do with reasoning ability.
What working memory is
Working memory is the small amount of information you can actively hold and manipulate at once. Not what you know — what you are currently juggling. It is famously limited, commonly described as around four items for most people, and it varies a good deal between individuals.
Now count what a three-rule matrix puzzle asks you to hold simultaneously: the count rule, the shape rule, the shading rule, which cell you are checking, what you have already verified, and the five answer options. That is comfortably past the limit for almost anyone.
So the puzzle does not fail because the reasoning is beyond you. It fails because you ran out of room to hold the reasoning in.
The fix is not to train memory. It is to stop needing it
Working memory capacity is quite stubborn and does not improve much with practice. What improves enormously is how much of it a given task demands — and that is entirely under your control.
Write the rules down
Four words in the margin — 'count +1, shape same' — moves a rule out of your head and onto paper, where it costs nothing to hold and cannot be lost. Every strong solver does this. The belief that doing it mentally is more impressive is the single most expensive habit in the whole subject.
Take one attribute at a time
Checking one rule against a grid uses a fraction of the capacity that checking three does. Three sequential passes are easier and faster than one simultaneous pass, even though there are three of them. This is the same technique as the stacked-rules method, and this is why it works — it is a working-memory intervention, not a reasoning one.
Use the page
Turn the paper to test a rotation. Cover the answer options with your hand so they stop competing for attention. Point at the cell you are on. All of these feel like crutches and all of them are simply moving load out of your head into the environment, which is what every expert in every field does.
This is why timed practice too early backfires. Time pressure consumes working memory on its own — part of your capacity goes to monitoring the clock. Practising slowly first builds rule recognition, and recognised rules cost almost nothing to hold. Speed comes from needing less room, not from having more.
Recognition is the real upgrade
When you first meet a progression rule you have to derive it: notice the counts, compare them, work out the step, hold it. That is several items of load.
After you have named that rule thirty times, you see the grid and think 'progression on count'. One item. The reasoning is identical; the cost has collapsed. That is the entire mechanism behind getting faster, and it is why naming rules matters so much more than doing volume.
If you find these harder than other people seem to
Working memory varies genuinely between individuals, and it is not a measure of intelligence or of how well you will do at anything in particular. Some people can hold six things and some four, and the four-item person who writes things down will outperform the six-item person who does not.
The techniques on this page are not a way of compensating for a deficiency. They are what the task actually requires, and the people who look effortless are mostly people who either have unusually low load because of practice, or who quietly externalise more than you notice.